Any new word on those optimized clients once mentioned?

ben
ben
Joined: 9 Feb 05
Posts: 36
Credit: 1,663
RAC: 0
Topic 189186

In one of the threads, Dr. Allen said there was work afoot by some volunteers to optimized the Einstein client. (SIMD, compiler optimizations, etc.)

Has there been any futher word on this?

Blank Reg
Blank Reg
Joined: 18 Jan 05
Posts: 228
Credit: 40,599
RAC: 0

Any new word on those optimized clients once mentioned?

Do not think it will happen cause they will not release the source......

Tom Merrick
Tom Merrick
Joined: 18 Jan 05
Posts: 1
Credit: 3,053,127
RAC: 0

That is a shame. With the

That is a shame. With the SETI optimized client I am doubling my throughput. Maybe Einstein cannot handle the load.

The Pirate
The Pirate
Joined: 11 Nov 04
Posts: 57
Credit: 23,332,769
RAC: 0

During beta testing of E@H,

During beta testing of E@H, before it was released to the public, my AMD64 3000 could do a wu in about 30 minutes. There was a post stating that to reduce the load on the server the size of the wu's were going to be increased about 12 times. Now my AMD64 3000 takes around 6 hours, 12 times longer on the button. I believe that with the longer wu's you are still doing the same amount of science as you would if you crunched 12 smaller wu's. I haven't checked but I assume that the credit you receive for the longer wu's went up by the same amount. So it's an even wash.


Divide Overflow
Divide Overflow
Joined: 9 Feb 05
Posts: 91
Credit: 183,220
RAC: 0

That's true, but I think that

That's true, but I think that there are crunchers here who are hopeful that Einstein@Home can see some of the dramatic speed improvements that Seti@Home has seen recently with some custom compiled applications. If that means increasing the size of the WU again because the completion times are too fast for the server loads, that would be great. At least we'd all be doing more efficient crunching for science.

Shaktai
Shaktai
Joined: 8 Nov 04
Posts: 183
Credit: 426,451
RAC: 0

It was quite a while back,

It was quite a while back, and I can't remember when, but there was some optimization work by the project staff on the Einstein apps for Linux and Macintosh, and possibly also on Windows. Those efforts did result in reduced crunch times especially for Linux and Macintosh which were at a significant disadvantage compared to the Windows apps, so the efforts have been made to optimize the Einstein apps to the maximum degree reasonable for each platform.

Currently, the science code changes so frequently, that it really isn't practical to try and maintain maximum optimization for all platforms, because of the amount of work required. I believe the project is striving for the best average optimization their compilers can provide for each platform, while maintaining stability and good science.

Rapidily changing science code and the need to maintain stability and accuracy are also reasons why the source code is not made public for einstein and some other projects. SETI has a relatively stable science code compared to the other projects, so it is more feasible there.

ML1
ML1
Joined: 20 Feb 05
Posts: 347
Credit: 86,483,719
RAC: 2,824

Rapidily changing science

Message 11694 in response to message 11693

Rapidily changing science code and the need to maintain stability and accuracy are also reasons why the source code is not made public for einstein and some other projects. SETI has a relatively stable science code compared to the other projects, so it is more feasible there.
This is precisely where making the source available for others to scrutinise and peer review will IMPROVE "stability and accuracy".

There is always an initial embarrassment to reveal the code, hacks - kludges - warts 'n' all. (Just clean up the worst of the 'frustration' comments!) The returns in the positive feedback from interested programmers will more than give positive compensation.

The natural peer review process should also speed the development process.

(Note how s@h has benefited from the Open Source ideals.)

Regards,
Martin

See new freedom: Mageia Linux
Take a look for yourself: Linux Format
The Future is what We all make IT (GPLv3)

arcturus
arcturus
Joined: 11 Feb 05
Posts: 44
Credit: 1,008,160
RAC: 0

Shaktai states: It was quite

Message 11695 in response to message 11693


Shaktai states:
It was quite a while back, and I can't remember when, but there was some optimization work by the project staff on the Einstein apps for Linux and Macintosh, and possibly also on Windows. Those efforts did result in reduced crunch times especially for Linux and Macintosh which were at a significant disadvantage compared to the Windows apps, so the efforts have been made to optimize the Einstein apps to the maximum degree reasonable for each platform.

Really?

On February 22 of this year I had posted that the same box taking 5 1/2 hrs to complete under Windows took over 9 1/2 with Linux.

Were these optimizations done after Feb 22nd or sometime beforehand to make these findings simply disappointing rather than pitiful? Link please.

Comment viewing options

Select your preferred way to display the comments and click "Save settings" to activate your changes.